Generates synthetic datasets for training and evaluating vision models on symbolic matching and navigation tasks. Each sample contains a maze where an agent must find a specific colored key and navigate to the matching colored door.
Each sample pairs a task (first frame + prompt describing what needs to happen) with its ground truth solution (final frame showing the result + video demonstrating how to achieve it). This structure enables both model evaluation and training.
| Property | Value |
|---|---|
| Task ID | G-45 |
| Task | Key Door Matching |
| Category | Spatiality |
| Resolution | 1024×1024 px |
| FPS | 16 fps |
| Duration | ~6-7 seconds |
| Output | PNG images + MP4 video |
# 1. Clone the repository
git clone https://github.com/VBVR-DataFactory/G-45_key_door_matching_data-generator.git
cd G-45_key_door_matching_data-generator
# 2. Create and activate virtual environment
python3 -m venv venv
source venv/bin/activate # On Windows: venv\Scripts\activate
# 3. Install dependencies
pip install --upgrade pip
pip install -r requirements.txt
pip install -e .# Generate 50 samples
python examples/generate.py --num-samples 50
# Custom output directory
python examples/generate.py --num-samples 100 --output data/my_dataset
# Reproducible generation with seed
python examples/generate.py --num-samples 50 --seed 42
# Without videos (faster)
python examples/generate.py --num-samples 50 --no-videos| Argument | Description |
|---|---|
--num-samples |
Number of tasks to generate (required) |
--output |
Output directory (default: data/questions) |
--seed |
Random seed for reproducibility |
--no-videos |
Skip video generation (images only) |
The scene shows a maze with a green circular agent, colored diamond-shaped keys, and colored hollow rectangular doors. Find the Red key and then navigate to the matching Red door, showing the complete movement process step by step.
 |
 |
 |
| Initial Frame Agent in maze with keys and doors |
Animation Agent finds key then navigates to matching door |
Final Frame Agent reaches correct colored door |
Navigate a maze to find a specific colored key, then proceed to the matching colored door.
- Agent: Green circular navigator
- Keys: Colored diamond shapes (various colors)
- Doors: Colored hollow rectangles (matching key colors)
- Maze structure: Walls creating pathways
- Target color: Specified in prompt (e.g., "Yellow")
- Two-phase task: (1) Find key, (2) Navigate to matching door
- Goal: Reach the door that matches the specified key color
- Color-based symbolic matching
- Two-step navigation task (key first, then door)
- Maze pathfinding with obstacles
- Tests color recognition and matching logic
- Requires planning multi-stage navigation
- Step-by-step movement visualization
data/questions/key_door_matching_task/key_door_matching_00000000/
├── first_frame.png # Maze with agent, keys, and doors
├── final_frame.png # Agent at matching door
├── prompt.txt # Key-door matching instruction with color
├── ground_truth.mp4 # Animation of key collection and door navigation
└── question_metadata.json # Task metadata
File specifications:
- Images: 1024×1024 PNG format
- Video: MP4 format, 16 fps
- Duration: ~6-7 seconds
logic-symbols color-matching maze-navigation multi-step-task symbolic-reasoning pathfinding